From: Greg Farnum <gfarnum@redhat.com>
Date: Mon, 10 Jul 2017 20:33:26 +0000 (-0700)
Subject: osd: do not default-abort on leaked pg refs
X-Git-Tag: v11.2.1~21^2
X-Git-Url: http://git-server-git.apps.pok.os.sepia.ceph.com/?a=commitdiff_plain;h=refs%2Fpull%2F14597%2Fhead;p=ceph.git

osd: do not default-abort on leaked pg refs

Signed-off-by: Greg Farnum <gfarnum@redhat.com>
(cherry picked from commit 4caf2df0c380a1281db9509b3feb342705512b58)
---

diff --git a/qa/clusters/extra-client.yaml b/qa/clusters/extra-client.yaml
index 349439cb331..107ad827135 100644
--- a/qa/clusters/extra-client.yaml
+++ b/qa/clusters/extra-client.yaml
@@ -7,3 +7,8 @@ openstack:
 - volumes: # attached to each instance
     count: 3
     size: 10 # GB
+overrides:
+  ceph:
+    conf:
+      osd:
+        osd shutdown pgref assert: true
diff --git a/qa/clusters/fixed-1.yaml b/qa/clusters/fixed-1.yaml
index 37686073308..1529378001f 100644
--- a/qa/clusters/fixed-1.yaml
+++ b/qa/clusters/fixed-1.yaml
@@ -6,5 +6,9 @@ overrides:
         osd crush chooseleaf type: 0
         osd pool default pg num:  128
         osd pool default pgp num:  128
+  ceph:
+    conf:
+      osd:
+        osd shutdown pgref assert: true
 roles:
 - [mon.a, osd.0, osd.1, osd.2, client.0]
diff --git a/qa/clusters/fixed-2.yaml b/qa/clusters/fixed-2.yaml
index 861e5091931..e99283787f4 100644
--- a/qa/clusters/fixed-2.yaml
+++ b/qa/clusters/fixed-2.yaml
@@ -5,3 +5,8 @@ openstack:
 - volumes: # attached to each instance
     count: 3
     size: 10 # GB
+overrides:
+  ceph:
+    conf:
+      osd:
+        osd shutdown pgref assert: true
diff --git a/qa/clusters/fixed-3-cephfs.yaml b/qa/clusters/fixed-3-cephfs.yaml
index 499c84c3718..c6573b43ee8 100644
--- a/qa/clusters/fixed-3-cephfs.yaml
+++ b/qa/clusters/fixed-3-cephfs.yaml
@@ -9,3 +9,8 @@ openstack:
 log-rotate:
   ceph-mds: 10G
   ceph-osd: 10G
+overrides:
+  ceph:
+    conf:
+      osd:
+        osd shutdown pgref assert: true
diff --git a/qa/clusters/fixed-3.yaml b/qa/clusters/fixed-3.yaml
index 8e622d2db94..668e8b374da 100644
--- a/qa/clusters/fixed-3.yaml
+++ b/qa/clusters/fixed-3.yaml
@@ -6,3 +6,8 @@ openstack:
 - volumes: # attached to each instance
     count: 3
     size: 10 # GB
+overrides:
+  ceph:
+    conf:
+      osd:
+        osd shutdown pgref assert: true
diff --git a/qa/clusters/fixed-4.yaml b/qa/clusters/fixed-4.yaml
index 7f051dc1fed..f4e4b76ec87 100644
--- a/qa/clusters/fixed-4.yaml
+++ b/qa/clusters/fixed-4.yaml
@@ -3,3 +3,8 @@ roles:
 - [mon.b, osd.1, osd.5, osd.9, osd.13] 
 - [mon.c, osd.2, osd.6, osd.10, osd.14] 
 - [osd.3, osd.7, osd.11, osd.15, client.0] 
+overrides:
+  ceph:
+    conf:
+      osd:
+        osd shutdown pgref assert: true
diff --git a/src/common/config_opts.h b/src/common/config_opts.h
index 70ee16b433c..3fa77dccc25 100644
--- a/src/common/config_opts.h
+++ b/src/common/config_opts.h
@@ -931,6 +931,7 @@ OPTION(osd_recovery_op_warn_multiple, OPT_U32, 16)
 
 // Max time to wait between notifying mon of shutdown and shutting down
 OPTION(osd_mon_shutdown_timeout, OPT_DOUBLE, 5)
+OPTION(osd_shutdown_pgref_assert, OPT_BOOL, false) // crash if the OSD has stray PG refs on shutdown
 
 OPTION(osd_max_object_size, OPT_U64, 100*1024L*1024L*1024L) // OSD's maximum object size
 OPTION(osd_max_object_name_len, OPT_U32, 2048) // max rados object name len
diff --git a/src/osd/OSD.cc b/src/osd/OSD.cc
index 7ac7396686b..a8004a5ae45 100644
--- a/src/osd/OSD.cc
+++ b/src/osd/OSD.cc
@@ -2872,7 +2872,9 @@ int OSD::shutdown()
 #ifdef PG_DEBUG_REFS
 	p->second->dump_live_ids();
 #endif
-        ceph_abort();
+	if (cct->_conf->osd_shutdown_pgref_assert) {
+	  ceph_abort();
+	}
       }
       p->second->unlock();
       p->second->put("PGMap");